Skip to content

New research, once a week

The week's new whitepapers in one short email.

You can stop at any time on the Opt-Out page.

A p-Value of 1.25 and a Negative Probability: SciGround Team Uncovers Errors in R and Julia

A statistical test returned a p-value of 1.25. Another returned a negative probability. Licklider, Inc., the Tokyo company behind SciGround, today detailed four reports that contributed to improvements in research…

A p-Value of 1.25 and a Negative Probability: SciGround Team Uncovers Errors in R and Julia

Licklider, Inc.17 November 20262 min read

R's official release credits the founder's reports; further corrections span R, Julia and Python tools.

TOKYO, Nov. 17, 2026 /PRNewswire/ -- A statistical test returned a p-value of 1.25. Another returned a negative probability. Licklider, Inc., the Tokyo company behind SciGround, today detailed four reports that contributed to improvements in research software. R's official 4.6.2 release credits founder Tasuku Kobayashi's reports and Ahmed Sattar Jabbar's suggestion for a Wilcoxon correction. R Core has now fixed, in its development version, a second reported issue involving its exact Kendall test.

The work supports mathematical verification for AI-assisted research.

R: a negative probability hidden in plain sight

For 20 perfectly concordant pairs, R's exact Kendall test returned approximately -4.44e-16 instead of 8.22e-19. Its standard summary displayed "p-value < 2.2e-16," hiding the negative value passed to downstream code. Taking its logarithm produces NaN.

R Core confirmed the fix in development revision c90891 under report #19319. R 4.6.2 remains affected by this Kendall issue, although it includes the separate Wilcoxon correction. These examples concern numerical accuracy, not reversals of a 5% significance decision.

For the Wilcoxon correction, users of R 4.6.0 and 4.6.1 should update to 4.6.2. R's announcement: https://stat.ethz.ch/pipermail/r-announce/2026/000729.html

Julia: three observations, an invalid p-value

HypothesisTests.jl v0.11.8 returned approximately 1.25 for a signed-rank test on [1, 2, -3]. Enumerating all eight sign assignments established the correct result: 1.0. The project confirmed resolution through a separately authored correction released in v0.12.0.

Python: checking the error bound itself

In python-flint, a particular constructor input could discard uncertainty for an ordinary decimal such as 0.1, producing an interval that excluded the intended value. The project merged repair #449. The public report documents affected versions and a workaround pending release.

"A calculation can finish successfully and still return an invalid result," said Kobayashi. "We independently establish the answer and take reproducible evidence to maintainers, helping improve software other researchers use."

Licklider credits the maintainers and contributors who implemented and reviewed the corrections. Reproduction details and upstream records: https://www.licklider.ai/engineering/

About Licklider

Founded in Tokyo in 2024, Licklider develops SciGround, mathematical verification and evidence management connecting research design, data, analysis and claims.

Media Contact
Tasuku Kobayashi, Founder & CEO
427094@email4pr.com
+81-80-7482-7670
https://www.licklider.ai/

Cision View original content to download multimedia:https://www.prnewswire.com/news-releases/a-p-value-of-1-25-and-a-negative-probability-sciground-team-uncovers-errors-in-r-and-julia-302941375.html

SOURCE Licklider, Inc.

Found this useful? Share it

New research, once a week

The week's new whitepapers in one short email.